BUG: Series inconsistent with overflow errors in Timedelta calcs

[venaturum]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

>>> x = pd.Series([pd.Timedelta(100000, "days"), pd.Timedelta(100000, "days")])

# Results in ValueError
>>> x.sum()

# Results in OverflowError
>>> pd.Timedelta(100000, "days")*2

# Results in OverflowError
>>> x + x

# Permits overflow
>>> x*2
0   -13504 days +00:25:26.290448384
1   -13504 days +00:25:26.290448384
dtype: timedelta64[ns]

Problem description

It would seem that pandas.Series calcs with Timedeltas are inconsistent in raising overflow errors, or not.
The exact error is also inconsistent (OverflowError vs ValueError).

Expected Output

Ideally x*2 in above example raises an error, and all overflow related errors are of type OverflowError.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : 5f648bf
python : 3.7.5.final.0
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.19041
machine : AMD64
processor : Intel64 Family 6 Model 158 Stepping 10, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : None.None

pandas : 1.3.2
numpy : 1.21.2
pytz : 2021.1
dateutil : 2.8.2
pip : 19.2.3
setuptools : 41.2.0
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : None
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[rush2406]

for x.sum() case, if you use x = x.dt.days before doing sum, it works

[venaturum]

for x.sum() case, if you use x = x.dt.days before doing sum, it works

Thanks, not quite the point though.

[jbrockmendel]

xref numpy/numpy#16352 (comment)

[venaturum]

For anybody with similar problems playing at home, if you don't care about aligned indexes, or you can guarantee the indexes align, then the following can be used as a workaround for element-wise multiplication (which generalises the problem) between two vectors

X  # Series of timestamps
Y  # Series of floats

pd.Series([x*y for x,y in zip(X,Y)])

It will produce a series of Timedeltas and throw overflow errors if the result of any multiplication is too big